LockedIn
OKaraoke
OKaraoke (Open Karaoke) is free and open source karaoke software to have fun with friends. It converts a YouTube URL into a full karaoke experience with pitch and lyrics guide and a minimalistic UI.
This involves the following steps:
- Download the YouTube video as an audio and save it on the server
- Separate the vocals from the background music using a trained model from WhisperX
- Identify the artist and song name from the video title using Gemini-ai
- Acquiring the lyrics from Genius
- Aligning the lyrics with the vocals (lyrics guide)
Computing the pitches over time of the vocals (pitch guide) Comparing the users’ pitch to the pitch guide to score
difficulties
dependencies issues were something that took a lot of time. youtube was preventing us from downloading the audio, so we had to find some workarounds. getting the correct pitch from the user microphone was hard due to noise and other reasons.
Built With
- daisyui
- demucs
- fastapi
- geminiai
- lyricsgenius
- python
- tailwind
- torch
- typescript
- vite
- yt-dlp
Log in or sign up for Devpost to join the conversation.